Horton

Re: [POC] hash partitioning

Amul Sul <sulamul@gmail.com>

View thread View on postgr.es
From: amul sul <sulamul@gmail.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Yugo Nagata <nagata@sraoss.co.jp>, David Steele <david@pgmasters.net>, Greg Stark <stark@mit.edu>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2017-05-15T10:57:13Z
Lists: pgsql-hackers

Attachments

On Wed, May 10, 2017 at 10:13 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Wed, May 10, 2017 at 8:34 AM, Ashutosh Bapat
> <ashutosh.bapat@enterprisedb.com> wrote:
>> Hash partitioning will partition the data based on the hash value of the
>> partition key. Does that require collation? Should we throw an error/warning if
>> collation is specified in PARTITION BY clause?
>
> Collation is only relevant for ordering, not equality.  Since hash
> opclasses provide only equality, not ordering, it's not relevant here.
> I'm not sure whether we should error out if it's specified or just
> silently ignore it.  Maybe an ERROR is a good idea?  But not sure.
>
IMHO, we could simply have a WARNING, and ignore collation, thoughts?

Updated patches attached.

Regards,
Amul

Commits

  1. Refactor get_partition_for_tuple a bit.

  2. Adopt Bob Jenkins' improved hash function for hash_any(). This changes the

  3. Improve hash_any() to use word-wide fetches when hashing suitably aligned